Jensen 2011.

Study characteristics
Methods	Study design: RCT Study grouping: parallel group How were missing data handled? NR Randomisation ratio: 3:1 (3 intervention levels) Recruitment method: NR Sample size justification and outcome used: NR Sampling method: officially designated urban poor HHs in the Chinese provinces of Hunan and Gansu were randomly selected to participate using lists kept at the local offices of the Ministry of Civil Affairs. Study aim or objective: to determine whether food price subsidies result in improved nutrition in poor Chinese HHs. Study period: April–December 2006 Unit of allocation or exposure: HHs
Participants	Baseline characteristics Intervention or exposure Age: NR Place of residence: NR Sex: NR Ethnicity and language: NR Occupation: NR Education: NR SES: mean family size: 0.1 subsidy 2.8 (SD 1.2), 0.2 subsidy 2.9 (SD 1.2), 0.3 subsidy 2.7 (SD 1.1); mean expenditure per capita: 0.1 subsidy CNY 279 (SD 274), 0.2 subsidy CNY 249 (SD 267), 0.3 subsidy CNY 290 (SD 376) Social capital: NR Nutritional status: mean calories per capita: 0.1 subsidy 1758 (SD 570); 0.2 subsidy 1767 (SD 526), 0.3 subsidy 1752 (SD 569); mean protein per capita, g: 0.1 subsidy 47.8 (SD 17.0), 0.2 subsidy 47.8 (SD 17.8), 0.3 subsidy 48.2 (SD 17.8); mean mineral intake per capita relative to RDA: 0.1 subsidy 1.02 (SD 0.36), 0.2 subsidy 1.02 (SD 0.36), 0.3 subsidy 1.01 (SD 0.35); mean vitamin intake per capita relative to RDA: 0.1 subsidy 1.20 (SD 0.47), 0.2 subsidy 1.19 (SD 0.43), 0.3 subsidy 1.21 (SD 0.43) Morbidities: NR Concomitant or previous care: NR Control Age: NR Place of residence: NR Sex: NR Ethnicity and language: NR Occupation: NR Education: NR SES: mean family size: 2.9 (SD 1.2); mean expenditure per capita: CNY 259 (SD 255) Social capital: NR Nutritional status: mean calories per capita: 1752 (SD 565); mean protein per capita, g: 48.5 (SD 19.4); mean mineral intake per capita relative to RDA: 1.00 (SD 0.34); mean vitamin intake per capita relative to RDA: 1.17 (SD 0.38) Morbidities: NR Concomitant or previous care: NR Overall: NR Inclusion criteria: officially designated as urban poor (HHs fell below a locally defined poverty threshold (the Di Bao line), typically CNY 100–200 per person per month or USD 0.41–0.82 per person per day, which is below even the World Bank's 'extreme' poverty line of USD 1 per person per day); located in 2 Chinese provinces which provide subsidies for staples goods (rice in Hunan and wheat flour in Gansu). Exclusion criteria: NR Pretreatment: no group differences for the total sample, but in the Hunan subsample the 0.3 subsidy group had higher vitamin and mineral intake compared to control as well as a smaller family size compared to the 0.2 subsidy group. In the Gansu subsample 0.1 and 0.2 subsidy groups both had a smaller family size compared to control, while the 0.3 subsidy group had lower protein per capita compared to control and lower mineral intake compared to the 0.2 subsidy group. Attrition per relevant group: NR. Total attrition between round 1 and 2 was < 1% (11/1300 HHs), and no HHs attrited between round 2 and 3. Description of subgroups measured and reported: baseline characteristics and outcomes were reported for Hunan and Gansu provinces as subgroups of the pooled data. Intervention group split into 3 levels: 0.1, 0.2 and 0.3 subsidy level (corresponding to CNY 0.1, CNY 0.2 and CNY 0.3 reduction per 500 g of staple good). Total number completed and analysed per relevant group: total of 1293 HHs (Hunan 644; Gansu 649) completed baseline surveys: 324 in control group and 324 for subsidy level 0.1, 324 for subsidy level 0.2 and 321 for subsidy level 0.3. 1271 HHs were included in the analysis, but no breakdown by control and intervention levels provided. Total number enrolled per relevant group: 1300 HHs enrolled. 324 HH completed baseline surveys in the control, 324 in subsidy level 0.1, 324 in subsidy level 0.2 and 321 in subsidy level 0.3. Total number randomised per relevant group: NR
Interventions	Intervention characteristics Intervention or exposure Food access intervention category: increase buying power Intervention type: food subsidy vouchers Description: vouchers entitling HHs to a subsidy in the price of the local staple good (rice in Hunan and wheat flour in Gansu) to 750 g per person per day (twice the mean per capita consumption as determined by preintervention study) that could be used immediately or accumulated and used when required within the intervention period. HHs in the treatment groups were given printed vouchers entitling them to a price reduction of CNY 0.10, CNY 0.20 or CNY 0.30 off the price of each 500 g of staple good. Subsidy stayed fixed for each HH over course of study. These subsidies represented substantial price changes, since the mean preintervention price of rice in Hunan was CNY 1.2 per 500 g, and the mean for wheat flour in Gansu was CNY 1.04 per 500 g. Duration of intervention period: June–October 2006 Frequency: vouchers printed in quantities of 1, 5 and 10 jin (500 g), and 1‐month supply of vouchers was distributed at start of each month, with each HH receiving vouchers for 750 g per person per day (about twice the mean per capita consumption as measured by the preintervention survey). All vouchers remained valid until the end of the intervention, giving HHs time to spend down any accumulated vouchers at end of study. Number of study contacts: baseline (April 2006) with 2 follow‐ups (September and December 2006) Providers: survey and intervention conducted by employees of provincial‐level agencies of Chinese National Bureau of Statistics. Delivery: printed food vouchers entitling intervention HHs to the subsidy corresponding with its intervention level. Vouchers were redeemable at local grain shops, the owners of which were later reimbursed for the cost of the vouchers and given a fixed payment for complying with the guidelines in implementing the subsidy. HHs could use the vouchers only to purchase the province‐specific staple good and were not permitted to resell the vouchers or the goods purchased with the vouchers (they were told there would be auditing and accounting to make sure they were in compliance with the rules, and that any violations would result in their removal from the study without any additional compensation). Co‐interventions: NR Resource requirements: NR Economic indicators: NR Control: no intervention
Outcomes	Adequacy of dietary intake: caloric/protein intake per capita Mineral Sufficiency index, Vitamin Sufficiency Index
Identification	Sponsorship source: National Institute of Aging, the William F. Milton Fund at Harvard Medical School, the Harvard Kennedy School's Dean's Research Fund, the Center for International Development at Harvard University, and the Hefner China Fund. Country: China Setting: extremely poor HHs in urban areas Authors' names: Robert T Jensen; Nolan H Miller Email: nmiller@illinois.edu; robertjensen@ucla.edu Declarations of interest: NR Study or programme name and acronym: N/A Type of record: journal article
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	Unclear risk	HHs were randomised to control or 3 levels of intervention, but no information provided on how this was achieved. The even distribution of HHs at baseline (324, 324, 324 and 321) indicated that a pseudo‐random technique may have been employed.
Allocation concealment (Selection bias)	Unclear risk	No information reported on whether or how the randomisation sequence was protected.
Baseline characteristics similar (Selection bias)	Unclear risk	Covariates were balanced across the entire pooled group, but significant differences existed at the province level. In Hunan, 0.3 subsidy HHs had significantly fewer members (t = 0.27; P < 0.05) when compared to 0.2 subsidy HHs; while in Gansu 0.1 and 0.2 subsidy HHs had significantly fewer HH members (0.1: t = 0.24; P < 0.05; 0.2: t = 0.19; P < 0.05) when compared to control.
Baseline outcome measurements similar (Selection bias)	Unclear risk	Overall, for the pooled sample randomisation appears to have achieved balance across the control and 3 treatment groups. Statistically significant differences exist between intervention and control groups at the province level, with 0.3 subsidy HHs in Hunan having significantly higher mineral (t = –0.074; P < 0.05) and vitamin (t = –0.13; P < 0.001) intake; and 0.3 subsidy HHs in Gansu having significantly lower (t = 3.84; P < 0.05) per capita protein consumption. In the latter province, 0.3 subsidy HHs also had a significantly lower mineral intake when compared to 0.2 subsidy HHs (t = 0.058; P < 0.05).
Blinding of participants and personnel (Performance bias)	Low risk	Given the nature of the intervention, it is very unlikely that participants could have been blinded to their allocation but this is unlikely to have introduced performance bias.
Blinding of outcome assessment (Detection bias)	High risk	As participants were the outcome assessors during the survey and were likely aware of their allocation to control or intervention, it is possible that this knowledge may have influenced the results.
Protection against contamination (Performance bias)	Low risk	Quote: "The possibility that the subsidy may attract other non eligible family members to the household is one case where the subsidy as we implemented it may yield different impacts from general subsidy. Our subsidy was assigned to only a subset of households, creating a potential pool of ineligible persons related to an eligible person. In the case of a universal subsidy for which all individuals are eligible, or a subsidy targeted to the poor where there is high correlation in poverty among relatives, we would not expect the same household composition response. While this is a potential threat to the external validity of our study, the fact that we find that no such changes took place makes this concern less important."
Incomplete outcome data (Attrition bias)	Low risk	Attrition was very low across the study, with < 1% of HHs (11/1300) being LTFU.
Selective outcome reporting (Reporting bias)	Unclear risk	No protocol available, but it appeared that all relevant outcomes in the methods section were reported in the results section.
Other bias	Low risk	None identified.